Using Multiple Data Stores in the Cloud: Challenges and Solutions - Data Management in Cloud, Grid and P2P Systems - page 92

- Domain object : This element allows to associate the client's ownership with
stored objects.
- Capability object :The capability objects describe the container 's capabilities
in order to discover the cloud capabilities in term of data storage.
This international standard support several features. Indeed, it enables the avail-
able capabilities discovery in the cloud storage offering. In addition, it supports
the containers and their contents management. It defines also meta-data to be
associated with containers and the objects they contain. So, CDMI allows to
define the data stores requirements which is R 12 but at a lower level (it is more
infrastructure oriented than platform oriented).
3.3 Data Management Contract
Truong et al. [11,12,13] propose to model and specify data concerns in data con-
tracts to support concern-aware data selection and utilization in Service as a
Service (SaaS). For this purpose, they define an abstract model to specify a data
contract using terms such as data rights, pricing model, etc. Moreover, they pro-
pose some algorithms and techniques in order to enforce the data contract usage.
In fact, they present a data contract compatibility evaluation algorithm. This
latter allows the combination of multiple data assets (Data mashup). In addi-
tion, they introduce a model (called DEscription MOdel for DaaS or DEMODS
[12]) for exchanging data agreements based on a new type of services which is
referred to as Data Agreement Exchange as a Service (DAES) [11]. However,
Truong et al. propose this data contract for service selection and composition
and not to store data or to help the developer to choose the appropriate data
stores for his application.
Ruiz-Alvarez et al. [14] propose an automated approach to select the PaaS
storage services according to a given application requirements. For this purpose,
they define an XML schema based on a machine readable description of the ca-
pabilities of each storage system. This XML schema allows consumers to express
their storage needs using high-level concepts. In addition, it is used to match
consumer's requirements and the data storage capabilities.
In summary, the current works on Data Management Contracts (DMCs), often
referred to as data agreements or data licenses, are not tailored to our purpose
which is to allow an application to bind to the appropriate data stores according
to its requirements. A possible solution is to use DMC to implement the data
storage offerings and the application needs matching that is the requirement
R 13 . CDMI and CAMP are interesting inputs in this direction.
4 Multiple Data Stores Requirements
In this section we analyze the requirements associated to the multiple data stores
scenario S 3 and describe the state of the art. It corresponds to the polyglot per-
sistence case [15]. Three requirements have been identified: ( R 3 ) the application